3d Welding

8/13/2019 3d Welding

1/11

RAPID PROTOTYPING USING D WELDINGr P M Dickens, r M S Pridham, Dr R C Cobb, Dr I Gibson and r G Dixon

Department of Manufacturing Engineering and Operations Management,University of Nottingham, UK.

IntroductionRapid prototyping systems are based, almost exclusively on polymer, or paper materials.The dimensions of the parts produced are limited by the volume of the processing area withinthe machine, and parts tend to warp or distort due to shrinkage and lack of support. Also themechanical properties of the part are restricted to those of the processable materials and thus,in many cases, required engineering properties cannot be obtained.Various European organisations are undertaking research to produce metal prototypesdirectly, for example, one project is concerned with a version of Laminated Object

Manufacturing where sheets of steel are laser welded together. There are also a number oforganisations using laser sintering to produce parts in Mild Steel and Stainless Steel.At Nottingham, work on Rapid Prototyping systems, based on a 3-D welding system,has been undertaken to try to combat some of the weaknesses of the other processes. There issome history of the use of welding as a means of building up components and parts for salvageand reclamation; thicknesses up to 50mm being typical 1). The use of welding for creating freestanding shapes was established in Germany in the 1960 s 2). This led to companies such asKrupp, Thyssen and Sulzer developing welding techniques for the fabrication of largecomponents of simple geometry, such as pressure vessels which could weigh up to 500 tonnes 3). The -technique was to become known as shape welding .Other work in this area has been undertaken by Babcock and Wilcox 4) who have beenworking mainly on large components produced in austenitic material. Also, work by RollsRoyce 5) has centred on investigating the technique s a means of reducing the wastage levelsof expensive high performance alloys which can occur in conventional processing. They havesuccessfully produced various aircraft engine parts n Nickel based and titanium based alloys.

As a production technique, 3-D welding offers significant advantages over conventionalprocessing, these include:The potential for robot control of the welding torch allowing large variation in partdimensions and geometry.A highly automated system.Parts with consistent properties.Rapid processing times, hence vastly reduced development times.Efficient use of materials.Direct production of a metal part - unique amongst current Rapid Prototyping.

8


2/11

3-D Welding TrialsInitial work on 3-D welding, verified the potential of welding as a Rapid Prototypingsystem. The first parts produced by robot welding were simple unsupported vertical walls ina square box formation figure 1 . The successful completion of these parts led to trials

involving the production of sloping walls in the form of a truncated pyramid figure 2 .Once it was established that more complicated structures were possible, a thermostathousing for a Ford automobile was obtained and the robot programmed to manufacture thisshape. Firstly the base and body of the housing was produced and then rotated through 90 toweld the outlet see figure 3 , this required 22 minutes of welding. This proved the ability tomanufacture parts that would normally be produced by casting. A sheet metal part was chosennext and figure 4 illustrates the original part with the welded version alongside. As there wasmore material in this part, the welding time was longer, at approximately 3 hours.Effect of Welding Parameters on Weld Bead Characteristics

The shape and dimensions of the weld bead are very important in the use of 3-D weldingas a Rapid Prototyping system, since these will determine the limits to the wall thickness whichmay be produced and will also influence the quality of the surface finish.Numerous trials were undertaken to produce single weld beads for a range of weldingconditions. The parameters which were varied included voltage, wire feed rate, wire stickout,wire diameter and welding velocity. The arc voltage and welding current under each conditionwas monitored and the dimensions of the bead produced height and width were subsequentlymeasured using a shadowgraph technique. A vast amount of data was generated in this way,but the general trends are presented in Table

Increasing Variable Effect on Measured VariableArc Voltage Current Bead Width Bead Height

Voltage t t tWire Feed t t tStickout t tWire Diameter t t tVelocity

Table 1 - Etlect on Measured Vanall sThe shape of the weld bead can be controlled to change from a wide-flat bead to amore narrow bead with vertical walls see figure 5 . When thin walls are to be produced thenthe best results are achieved by using a narrow bead, however, there are many situations wherebeads will need to be placed along side each other and so a variety of bead shapes could beused, depending on the slope of the wall surface see figure 6 .This work has produced information to start constructing a welding database but furtherwork is required, areas such as the effect of multiple layers on weld bead dimensions, requirefurther investigation, even though wall smoothness has improved see figure 7 .

281


3/11

MaterialsAs one the main advantages this technique is the direct production a metal part,it is very important to confirm the mechanical and structural properties parts produced in thisway. To do this, a programme mechanical and microstructural examination was undertaken.

Using typical operating parameters and mild steel welding wire .based on Fe-C 0.08 ) Si: O.9 ) Mn 1.5 ), square box section walls were produced and test-pieces were cut atdifferent positions and orientations within the walls.Vickers hardness measurements 10kg load) made in a range positions over the wallsurface, showed little variation along the wall length, but a definite increase in hardness from146.3 VHN at the base to 172.6 VHN at the top the wall was noted the height the wallbeing approximately l00mm and corresponding to around 70 welding passes).Tensile tests were carried out on test-pieces cut from various orientations in the wall,with samples being taken:-i) With the tensile axis corresponding to the vertical direction andii) With the tensile axis corresponding to the horizontal direction, both from theupper and lower part the wall.Results were consistant, regardless orientation, with typical values 490 MPa measured.. Values obtained from vertical specimens indicated few areas weakness betweenadjacent weld layers. Elongation to failure values however, showed a marked differencebetween horizontal test-pieces cut from the bottom and top the wall. Typical values for thebase the wall are in excess 30 whilst those at the top average 22.5 . The measuredmechanical properties are summarised in Table 2.

g

Specimen Orientation Ultimate Tensile Elongation to FailureStrength MPaVertical 489 35.0Horizontal Top) 484 22.5Horizontal Bottom) 499 33.1

Table 2 - Avera e TenSIle Test DataOptical microscopy revealed a microstructure in the wall largely equiaxed ferrite exand pearlite ex+ F ~ with a grain size approximately 60j.tm see figure 8. There are localregions where the structure is much less equiaxed and more columnar in nature grain sizeapproximately 600j.tm x l00j.tm) as would be expected in an as-cast material, seen in figure 9,These regions, however, are restricted to the very top layers the wall where the layers havenot been subjected to reheating, from the deposition subsequent weld layers. This accountsfor the lower levels ductility measured in specimens cut from the top the wall in thehorizontal direction. A simple heat treatment procedure could be used to produce a moreuniform microstructure throughout the wall.Microscopy revealed no voids or cavities and the measured density was approximately99.5 equivalent wrought material. No indication was found oxidation between adjacentlayers, this is supported by the strength measurements for the vertical test piecesResults indicate that parts produced by the 3-D welding process have good structuralintegrity and property levels which would allow them to be exploited in service conditions.


4/11

Off ine Robot ProgrammingThe parts which have been produced and described so far have been produced using acumbersome and highly time consuming on-line robot programming technique. is evident thatto be a viable process for producing prototype parts in the desired timescale, an efficient off-line

programming system is required.To test the robot welding system for more complex and representative shapes, a CADsystem was used to generate welding trajectories. The following procedure was adopted:- A CAD image the part to be manufactured was created using the PEPS2 CADmachining package 6 . This package does not have a robot post processor but theinstructions are similar to those used for milling. The milling process could beconsidered to be the negative the welding process.2 Once the part has been designed and a simulation the process has been successfullypresented, the post processor generates the spatial positions required for the welding

torch along with commands to signify the tool status torch/gas on/oft .3 The file positions is not fully understandable to the robot. A further post processingwritten on a PC in the C programming language is required to include robot commandwords. This file is then concatenated with an initialisation programme which providesstart positions, velocity profiles and signal levels. This concatenated file is then capable

being directly down-loaded to the robot for part fabrication.A number limitations to the system can be identified and some modifications need tobe made.As has already been mentioned, there is no feedback loop between the robot controller

and the welding system. is not possible, therefore, to dynamically control the welding qualitythrough the robot system. Sensory feedback would be a requirement for improvement thesystem quality through process monitoring and for post inspection purposes. Some thisinformation could be used to direct any further machining the part that may be required.Attention must also be given to the use sensors to prevent possible collapse the partthrough temperature build up. The CAD/CAM link and the post processor is only limitedvalue, allowing only sequential machine operations with no facility for sensory update. ThePEPS2 system will only provide code using a cartesian coordinate system for 3 axis machineswhich is only limited use for robot systems. The technique used required a substructure thatis directly below the current welding path. Part manipulation is therefore required to build upcomplex shapes like the thermostat housing in figure 3 by combining simpler ones. The needfor extra degrees freedom and integration robots with part manipulators rotary XY tablesrequires a software simulation and post processing system beyond the capabilities PEPS2.

8


5/11

uture evelopmentsThe 3D robot welding system relies as much on CAD/CAM technology as welding androbotic technology. Further development will be linked very strongly with the CAD systemused and the manipulator post processor. The fabrication procedure involves the building up

complex geometric shapes from more simple constituent shapes. has been found thatcylinders for instance are best built in the form rings laid on top each other. can beseen, therefore, that the layer formation used by conventional rapid prototyping is not whollyappropriate for this technique. Since the constituent shapes can be placed in an infinite number ways with respect to each other, it is obvious that some form part manipulation will berequired in conjunction with the robot welder. A future system must therefore incorporate botha robot and a 3-axis part manipulator.

The formulation the geometry will require the adaption an existing CAD package.This package will be used to design the part and then determine the best geometric proceduresrequired to fabricate it. This information will then be passed onto a robot simulation packageeg GRASP which will calculate the robot and part manipulator instructions. The postprocessor within this system will be used to programme the robot once simulation has beenshown to adequately construct the part. This whole procedure design and simulation mayrequire an iterative process with some human intervention.Initial experiments have already shown that, although complex shapes can be formed,the results are not perfect. There are a number basic reasons for such imperfections:-Heat build up due to the welding process can cause earlier welding passes to remelt andcause part distortion or collapse the structure.Inaccuracies in the welding and robot parameters can cause cumulative errors, resultingin the torch being too close or too far away from the surface.Solid layers ie filling in outline shapes cannot be performed sufficiently accuratelyto form a smooth surface. This means that gaps can occur inside solid objects. is evident from these problems that some form sensing is required to control theprocess. Real time weld monitoring systems do exist, but it is unlikely that such systems canbe used to overcome all these difficulties. However, it will be necessary to stop welding atintermittent stages to leave the part to cool down and avoid collapse due to heat build up, r toimplement forced cooling. Sensors can be used at this time to determine how much cooling is

required and to perform post inspection the welding process. This inspection information canbe used to update the robot position to avoid cumulative errors, plot new robot trajectories toavoid gaps and generally control the weld parameters.

8


6/11

ConclusionsThe object o this work has been to expand the use o 3 Dimensional Welding from largesimple pressure vessels to a wide range o part sizes having much greater geometricalcomplexity.There are several targets to aim for in the future:Part Complexity: Parts will be produced such s engine castings cylinder heads and inletand exhaust manifolds in order to demonstrate the capability o the system.Part accuracy: At present the part accuracy is about O.5mm but it is expected thateventually this will be reduced to O.2mm.Surface Roughness: The present poor surface roughness is unacceptable for manysituations but better weld control and appropriate cooling techniques will improve thisconsiderably.

The applications for this technique can be characterised by two main groups:Parts: Prototype parts or small batches can be rapidly produced without any specifictooling. In addition small modifications could be made to existing components suchas different brackets or lugs on a car chassis.Tooling: it is possible to build a part without tooling then it should also be possibleto produce prototype tools such s press or forge tools plastic molds and dies.

Further investigation o the welding technology will also be required. s yet otherwelding techniques that may result in smoother more predictable finishes have not been tried.Investigation into this branch o technology may result in the development o new welding basedprocesses. In the past it has usually been more important to develop new welding systems thatcan lay down as much weld as possible in a short time. The 3 D welding system relies onprecision methods and may require development o robotic welding system with very smalldiameter filler wires.

8


7/11

References American Society for Metals, 1983, Metals Handbook, 9th Edition, Vol Welding,Brazing and Soldering prepared under the direction the ASM Handbook Committee,Ohio, pp771-803.2. Kassmaul, K.; Schoch, F.W. and Lucknow, H., September 1983, High Quality LargeComponents Shape Welded by a SAW process , Welding Journal.3. Piehl, K-H., January 1989, Formgebendes Schweipen von Schwekomponenten ,Company report: Thyssen Aktiengesellschaft, Duisburg.4. Doyle, T.E. Babcock Wilcox), October 1991, Shape Melting Technology , TheThird International Conference on Desktop Manufacturing: Making Rapid PrototypingPay Back, The Management Round Table.5. Private Communication with the Manufacturing Technology Section,Rolls-RoyceAerospace Group, Bristol, U.K.6. Camtek, March 1990, PEPS2 NC Part Programming System, Version 2.3 , CamtekLtd.7. Sorenti, P. and Bennaton, J., December 1989, Off-line programming moves intopractice , The Industrial Robot, vol 16 4), pp 205-207.

286


8/11

Figure Square Box Formation

Figure 2 Truncated Hollow Pyramid

8


9/11

Figure Welded Thermostat Housing

Figure 4 Sheet Metal Part and Welded Version


10/11

Figure Cross section welded walls before and after investigating theeffect weld bead parameters 9


11/11

Figure 6 Large equiaxed Ferrite and Pearlite structure

Figure 7 As cast type structure

29

Documents

3d Welding